From: Jeff Liebermann Subject: Re: Any experience with doppler rdf (radio direction finders)? Newsgroups: rec.radio.amateur.misc Date: 1994-04-12 View: Complete Thread (9 articles) | Original Format 23:06:30 PST In article <1994Apr8.220021.29409@Csli.Stanford.EDU> pkahn@Csli.Stanford.EDU (Philip Kahn) writes: >I have been reading up on doppler RDF's. The Amateur Radio Handbook >has an article that says they can only do well to about 5 degrees. >Have you heard of systems or ways to do it that gives better results? In 1976, I helped design the AN/SRD-22 doppler direction finder for Intech. The USCG specification was 1 degree resolution. 3 degree total variation over a wider temperature range. This is radically better than the typical Roanoak or Doppler systems design. This accuracy was achievable, but in my estimation worthless. Reasons to follow. In order to maintain a stable phase shift from the antenna, through the receiver, and finally to the inevitable phase comparator, some compensation is required. The following are the basic error sources and the solutions required. The antenna is 4 vertical tubes with pin diode attenuators (not switches) in series. These attenuators are driven in quadrature with a modified sine wave to give a low distortion diode resistance vs time waveform. Any distortion in the drive waveform cause errors at points BETWEEN quadrature points. The largest cause of phase errors in the receiver is the IF crystal filter. The AN/SRD-22 uses a 4KHz drive signal where the 2nd harmonic and above (i.e. any distortion) ends up outside the IF crystal filter bandwidth. Slight changes in frequency yielded monsterous changes in phase shift through the filter. The receiver needed an automatic frequency control to insure that the ADF tones would always land on the same place in the IF xtal filter. With weak signals, it was found that signal strengh affected the phase shift through the receiver. The final limiter and quadrature demodulator was the largest contributor. An automatic gain control was required to fix this. The demodulated 4Khz tone was filtered after demodulation by a commutating filter. The initial bandwith was about 1Hz. Clock leakage and switching transients became a large source of errors. This required very carful board layout. To give a rough approximation of the distortion requirements, 1 degree total accuracy is 1 part in 360 or 0.28% total error or distortion. At 1Hz bandwidth, it may take 20 seconds to obtain a stable reading. The reason the Roanoak and Doppler Systems switched antenna systems function is that the environmental sources of error far exeed the instrument errors. A 130ft vessel or moving vehicle is NOT a stable platform and is subject to errors induced by ground or sea reflections, multipath, antenna tilt, polarization, faraday rotation, mechanical damage, and multiple transmitter sources. Any ONE of these can induce enormous errors. Multiple transmitters (including intermod, garbage, grunge, rfi, emi, computer noise, skip, etc) are the worst as they generate totally false readings. Another major botch is the 0-360degree digital display. Since the readings tend NOT to be stable, getting an accurate bearing is like reading a digital watch that can't decide what time it is. Error induced by drive waveform distortion, signal levels, and such are reasonably small. There are other problems introduced by switching type antennas, but they do not affect the basic accuracy. Only the crystal filter errors are large enough to justify an AFC. My guess is that +/-5 degree overall accuracy for a fixed base is the best that can be expected. +/-15 degrees is about the best for a mobile. This technology is now 20 year old. Were I to do it today, I would use DSP to filter and demodulate the tone, statistical algorithms to reduce erronious readings, beam steering to deal with reflections, adjustable filter bandwidth, and intelligent display control. -- # Jeff Liebermann Box 272 1540 Jackson Ave Ben Lomond CA 95005 # 408.336.2558 voice wb6ssy@ki6eh.#nocal.ca.usa wb6ssy.ampr.org [44.4.18.10] # 408.699.0483 digital_pager 73557,2074 cis [don't] # jeffl@comix.santa-cruz.ca.us scruz.ucsc.edu!comix!jeffl